A single-cell analysis of breast cancer cell lines to study tumour heterogeneity and drug response.

Cancer cells within a tumour have heterogeneous phenotypes and exhibit dynamic plasticity. How to evaluate such heterogeneity and its impact on outcome and drug response is still unclear. Here, we transcriptionally profile 35,276 individual cells from 32 breast cancer cell lines to yield a single cell atlas. We find high degree of heterogeneity in the expression of biomarkers. We then train a deconvolution algorithm on the atlas to determine cell line composition from bulk gene expression profiles of tumour biopsies, thus enabling cell line-based patient stratification. Finally, we link results from large-scale in vitro drug screening in cell lines to the single cell data to computationally predict drug responses starting from single-cell profiles. We find that transcriptional heterogeneity enables cells with differential drug sensitivity to co-exist in the same population. Our work provides a framework to determine tumour heterogeneity in terms of cell line composition and drug response.

PATRI, a Genomics Data Integration Tool for Biomarker Discovery.

The availability of genomic datasets in association with clinical, phenotypic, and drug sensitivity information represents an invaluable source for potential therapeutic applications, supporting the identification of new drug sensitivity biomarkers and pharmacological targets. Drug discovery and precision oncology can largely benefit from the integration of treatment molecular discriminants obtained from cell line models and clinical tumor samples; however this task demands comprehensive analysis approaches for the discovery of underlying data connections. Here we introduce PATRI (Platform for the Analysis of TRanslational Integrated data), a standalone tool accessible through a user-friendly graphical interface, conceived for the identification of treatment sensitivity biomarkers from user-provided genomics data, associated with information on sample characteristics. PATRI streamlines a translational analysis workflow: first, baseline genomics signatures are statistically identified, differentiating treatment sensitive from resistant preclinical models; then, these signatures are used for the prediction of treatment sensitivity in clinical samples, via random forest categorization of clinical genomics datasets and statistical evaluation of the relative phenotypic features. The same workflow can also be applied across distinct clinical datasets. The ease of use of the PATRI tool is illustrated with validation analysis examples, performed with sensitivity data for drug treatments with known molecular discriminants.

Thermodynamics of the high-affinity interaction of TCF4 with beta-catenin.

The formation of a complex between beta-catenin and members of the TCF/LEF family of high-mobility group proteins is a key regulatory event in the wnt-signaling pathway, essential for embryonal development as well as the growth of normal and malignant colon epithelium. We have characterized the binding of TCF4 to human beta-catenin by steady-state intrinsic fluorescence quenching experiments, surface plasmon resonance (SPR) and isothermal titration calorimetry (ITC). Binding studies in solution and in heterogeneous phase showed that TCF4 binds reversibly to beta-catenin with an affinity (KB) of 3(+/-1) 10(8) M(-1). Site-directed mutagenesis, together with calorimetric measurements, revealed that residue D16 in TCF4 plays a crucial role in high-affinity binding. Mutation of this residue to alanine resulted in a decrease of KB by two orders of magnitude as well as a significant reduction in binding enthalpy. Binding of TCF4 to beta-catenin gave rise to a large negative enthalpy change at 25 degrees C (-29.7 kcal/mol). Binding enthalpies were strongly temperature dependent, which resulted in the determination of a large heat capacity change upon binding of -1.5 kcal/(mol K). The molecular events that take place upon complex formation are discussed using the measured thermodynamic data together with the crystal structure of the beta-catenin arm repeat region/TCF complex.

Early Th1 response in unprimed nonobese diabetic mice to the tyrosine phosphatase-like insulinoma-associated protein 2, an autoantigen in type 1 diabetes.

The insulinoma-associated protein 2 (IA-2) is a phosphatase-like autoantigen inducing T and B cell responses associated with human insulin-dependent diabetes mellitus (IDDM). We now report that T cell responses to IA-2 can also be detected in the nonobese diabetic (NOD) mouse, a model of human IDDM. Cytokine secretion in response to purified mouse rIA-2, characterized by high IFN-gamma and relatively low IL-10 and IL-6 secretion, was elicited in spleen cells from unprimed NOD mice. Conversely, no response to IA-2 was induced in spleen cells from BALB/c, C57BL/6, or Biozzi AB/H mice that express, like NOD, the I-A(g7) class II molecule, but are not susceptible to spontaneous IDDM. The IA-2-induced IFN-gamma response in NOD spleen cells could already be detected at 3 wk and peaked at 8 wk of age, whereas the IL-10 secretion was maximal at 4 wk of age and then waned. IA-2-dependent IFN-gamma secretion was induced in CD4(+) cells from spleen as well as pancreatic and mesenteric lymph nodes. It required Ag presentation by I-A(g7) molecules and engagement of the CD4 coreceptor. Interestingly, cytokines were produced in the absence of cell proliferation and IL-2 secretion. The biological relevance of the response to IA-2 is indicated by the enhanced IDDM following a single injection of the recombinant protein emulsified in IFA into 18-day-old NOD mice. In addition, IFN-gamma production in response to IA-2 and IDDM acceleration could be induced by IL-12 administration to 12-day-old NOD mice. These results identify IA-2 as an early T cell-inducing autoantigen in the NOD mouse and indicate a role for the IA-2-induced Th1 cell response in IDDM pathogenesis.

Activation of Zap-70 tyrosine kinase due to a structural rearrangement induced by tyrosine phosphorylation and/or ITAM binding.

The protein tyrosine kinase ZAP-70 is implicated in the early steps of the T-cell antigen receptor (TCR) signaling. Binding of ZAP-70 to the phosphorylated immunoreceptor tyrosine-based activation motifs (ITAMs) of the TCR zeta chain through its two src-homology 2 (SH2) domains results in its activation coupled to phosphorylation on multiple tyrosine residues, mediated by Src kinases including Lck as well as by autophosphorylation. The mechanism of ZAP-70 activation following receptor binding is still not completely understood. Here we investigated the effect of intramolecular interactions and autophosphorylation by following the kinetics of recombinant ZAP-70 activation in a spectrophotometric substrate phosphorylation assay. Under these conditions, we observed a lag phase of several minutes before full ZAP-70 activation, which was not observed using a truncated form lacking the first 254 residues, suggesting that it might be due to an intramolecular interaction involving the interdomain A and SH2 region. Accordingly, the lag phase could be reproduced by testing the truncated form in the presence of recombinant SH2 domains and was abolished by the addition of diphosphorylated ITAM peptide. Preincubation with ATP or phosphorylation by Lck also abolished the lag phase and resulted in a more active enzyme. The same results were obtained using a ZAP-70 mutant lacking the interdomain B tyrosines. These findings are consistent with a mechanism in which ZAP-70 phosphorylation/autophosphorylation on tyrosine(s) other than 292, 315, and 319, as well as engagement of the SH2 domains by the phosphorylated TCR, can induce a conformational change leading to accelerated enzyme kinetics and higher catalytic efficiency.

Functional characterization and mechanism of action of recombinant human kynurenine 3-hydroxylase.

The mitochondrial outer membrane enzyme kynurenine 3-hydroxylase (K3H) is an NADPH-dependent flavin mono-oxygenase involved in the tryptophan pathway, where it catalyzes the hydroxylation of kynurenine. K3H was transiently expressed in COS-1 cells as a glutathione S-transferase (GST) fusion protein, and the pure recombinant protein (rec-K3H) was obtained with a specific activity of about 2000 nmol.min-1.mg-1. Rec-K3H was shown to have an optimum pH at 7.5, to use NADPH more efficiently than NADH, and to contain one molecule of non-covalently bound FAD per molecule of enzyme. The mechanism of the rec-K3H-catalyzed reaction was investigated by overall initial-rate measurements, and a random mechanism in which combination of the enzyme with one substrate does not influence its affinity for the other is proposed. Further kinetic studies revealed that K3H activity was inhibited by both pyridoxal phosphate and Cl-, and that NADPH-catalyzed oxidation occurred even in the absence of kynurenine if 3-hydroxykynurenine was present, suggesting an uncoupling effect of 3-hydroxykynurenine with peroxide formation. This observation could be of clinical interest, as peroxide formation could explain the neurotoxicity of 3-hydroxykynurenine in vivo.

Role of the Src homology 2 domains and interdomain regions in ZAP-70 phosphorylation and enzymatic activity.

The protein tyrosine kinase ZAP-70, which mediates T-cell antigen receptor (TCR) signalling, contains three distinct functional modules, two tandemly arranged SH2 domains, a kinase domain and a linker region (interdomain B) that connects them. ZAP-70 enzymatic activation is strictly dependent on the binding, via its SH2 domains, to the triggered TCR and on tyrosine phosphorylation. Here we utilized recombinant ZAP-70 and carried out a mutational analysis to understand the structural requirements for its activation. We show that deletion of both SH2 domains corresponding to the first 254 residues moderately increases ZAP-70 enzymatic activity on an exogenous substrate in vitro, results in increased tyrosine phosphorylation and produces subtle conformational changes, as judged by altered SDS/PAGE migration. Mutation of Tyr292, 315 and 319 to Phe in the interdomain B region, which constitute the major phosphorylation sites both in vitro and in vivo, did not affect ZAP-70 enzymatic activity. Moreover, deletion analysis of the interdomain B region established residues 320-619 as a minimal region endowed with full kinase activity. We propose that binding of ZAP-70 to the TCR promotes, through conformational changes, its extensive phosphorylation on tyrosine. However, Tyr292, 315 and 319 do not affect ZAP-70 enzymatic activity and may influence ZAP-70 signalling only indirectly by mediating its association with intracellular transducers.

Tyrosine 319, a newly identified phosphorylation site of ZAP-70, plays a critical role in T cell antigen receptor signaling.

Following T cell antigen receptor (TCR) engagement, the protein tyrosine kinase (PTK) ZAP-70 is rapidly phosphorylated on several tyrosine residues, presumably by two mechanisms: an autophosphorylation and a trans-phosphorylation by the Src-family PTK Lck. These events have been implicated in both positive and negative regulation of ZAP-70 activity and in coupling this PTK to downstream signaling pathways in T cells. We show here that Tyr315 and Tyr319 in the interdomain B of ZAP-70 are autophosphorylated in vitro and become phosphorylated in vivo upon TCR triggering. Moreover, by mutational analysis, we demonstrate that phosphorylation of Tyr319 is required for the positive regulation of ZAP-70 function. Indeed, overexpression in Jurkat cells and in a murine T cell hybridoma of a ZAP-70 mutant in which Tyr319 was replaced by phenylalanine (ZAP-70-Y319F) dramatically impaired anti-TCR-induced activation of the nuclear factor of activated T cells and interleukin-2 production, respectively. Surprisingly, an analogous mutation of Tyr315 had little or no effect. The inhibitory effect of ZAP-70-Y319F correlated with a substantial loss of its activation-induced tyrosine phosphorylation and up-regulation of catalytic activity, as well as with a decreased in vivo capacity to phosphorylate known ZAP-70 substrates, such as SLP-76 and LAT. Collectively, our data reveal the pivotal role of Tyr319 phosphorylation in the positive regulation of ZAP-70 and in TCR-mediated signaling.

Tyrosine 474 of ZAP-70 is required for association with the Shc adaptor and for T-cell antigen receptor-dependent gene activation.

The protein tyrosine kinase ZAP-70 plays a central role in T-cell activation. Following receptor engagement, ZAP-70 is recruited to the phosphorylated subunits of the T-cell antigen receptor (TCR). This event results in ZAP-70 activation and in association of ZAP-70 with a number of signaling proteins. Among these is the Shc adaptor, which couples the activated TCR to Ras. Shc interaction with ZAP-70 is mediated by the Shc PTB domain. The inhibitory effect of a Shc mutant containing the isolated PTB domain suggests that Shc interaction with ZAP-70 might be required for TCR signaling. Here, we show that a point mutation (Phe474) of the putative Shc binding site on ZAP-70, spanning tyrosine 474, prevented ZAP-70 interaction with Shc and the subsequent binding of Shc to phospho-zeta. Neither ZAP-70 catalytic activity nor the pattern of protein phosphorylation induced by TCR triggering was affected by this mutation. However expression of the Phe474 ZAP-70 mutant resulted in impaired TCR-dependent gene activation. ZAP-70 could effectively phosphorylate Shc in vitro. Only the CH domain, which contains the two Grb2 binding sites on Shc, was phosphorylated by ZAP-70. Both Grb2 binding sites were excellent substrates for ZAP-70. The data show that Tyr474 on ZAP-70 is required for TCR signaling and suggest that Shc association with ZAP-70 and the resulting phosphorylation of Shc might be an obligatory step in linking the activated TCR to the Ras pathway.

Substitution of two variant residues in the protein tyrosine phosphatase-like PTP35/IA-2 sequence reconstitutes catalytic activity.

The PTP35/IA-2 protein shows high homology to protein tyrosine phosphatases (PTPases) but harbours a few changes in invariant PTPase residues. Accordingly, PTP35/IA-2 has been reported to lack catalytic activity in vitro, and its in vivo biological function remains to be determined. We investigated if reversion of selected amino acids to the PTPase consensus could reconstitute enzymatic activity. Substitution of aspartic acid 911 in the putative active site with alanine resulted in the appearance of low but reproducible activity on pNPP dephosphorylation. Moreover, contemporary replacement of alanine 877 with aspartic acid greatly increased the catalytic efficiency of the D911A mutant. The A877D/D911A double mutant protein was also found to specifically dephosphorylate myelin basic protein phosphorylated on tyrosine. These results suggest that the general scaffold of the PTP35 protein is compatible with a common catalytic mechanism shared by PTPases and argue against an intrinsic enzymatic function of the wild type form.

Expression of PTP35, the murine homologue of the protein tyrosine phosphatase-related sequence IA-2, is regulated during cell growth and stimulated by mitogens in 3T3 fibroblasts.

Protein tyrosine phosphatases (PTPases) have been implicated in the control of cell proliferation and differentiation. To isolate new members of this family potentially involved in cell growth regulation, we looked for PTPase sequences differently expressed in proliferating or quiescent NIH 3T3 fibroblasts. The full-length cDNA of one of these growth-regulated genes, named PTP35, was isolated from a 3T3 library and found to encode the murine IA-2 PTPase-related sequence. Endogenous PTP35 mRNA steady-state levels were found to be strictly regulated during cell growth in 3T3 fibroblasts, being high in actively cycling cells and barely detectable in density-arrested cells. Both PTP35 mRNA and protein levels could be induced in quiescent cells by mitogenic stimulation. The growth factor specificity and kinetics of this induction were analyzed in detail.

Compositional compartmentalization of the nuclear genomes of Trypanosoma brucei and Trypanosoma equiperdum.

High molecular weight DNA preparations from Trypanosoma brucei and Trypanosoma equiperdum were fractionated by preparative centrifugation in a Cs2SO4 density gradient in the presence of BAMD, bis(acetatomercurimethyl)dioxane, a sequence-specific DNA ligand. Analytical centrifugation in CsCl of the DNA fractions so obtained showed that both DNAs had a bimodal distribution with two major peaks banding at 1.702-1.703 and 1.708 g/cm3 and representing 1/3 and 2/3 of total DNA, respectively. Several minor components were also detected. These results indicate that a compositional compartmentalization is not only found in the genome of vertebrates and plants, as already described, but also in those of protozoa such as Trypanosomes.

Biochemical characterization of the molecular interaction between recombinant basic fibroblast growth factor and a recombinant soluble fibroblast growth factor receptor.

The extracellular domain of human fibroblast growth factor receptor (XC-FGF-R) was expressed in Escherichia coli. The protein was purified to homogeneity and the interaction with basic fibroblast growth factor (bFGF), its physiological ligand, was examined. Using resins on which bFGF was reversibly bound, we analysed the characteristics of the binding between XC-FGF-R and immobilized bFGF. We also investigated the stoichiometry of the binding between XC-FGF-R and recombinant human bFGF (rhbFGF) applying non-denaturing gel electrophoresis, chemical cross-linking followed by SDS/PAGE, and gel-filtration chromatography. In cross-linking and gel-filtration chromatography experiments, a 1:1 complex between rhbFGF and XC-FGF-R was observed. The complex was separated from the non-complexed proteins using non-denaturing PAGE in the presence of 0.1% Triton X-100. The band corresponding to the complex was recognized by specific antibodies directed against bFGF and its receptor, blotted on poly(vinylidene difluoride) membranes and submitted to sequence and amino acid analysis. The data obtained from these determinations confirmed the formation of a 1:1 complex between rhbFGF and XC-FGF-R.

Biochemical bases of the interaction of human basic fibroblast growth factor with glycosaminoglycans. New insights from trypsin digestion studies.

In the present study we have attempted a characterization of the biochemical bases of the interaction of human basic fibroblast growth factor (bFGF) with glycosaminoglycans (GAGs) in solution. This interaction has been evidenced as the capacity of different GAGs and various sulfated compounds to protect bFGF and different bFGF mutants from tryptic cleavage. Heparin protects bFGF from trypsin digestion in a dose-dependent fashion. Substitution by site-directed mutagenesis of two or more basic residues with neutral glutamine residues in the amino-terminal region bFGF(27-32) or in the carboxyl-terminal region bFGF(118-129) does not significantly affect the protective effect exerted by heparin. In contrast, heparin protection is abolished when the full region bFGF(27-32) is deleted. The capacity of different GAGs to protect bFGF from proteolytic cleavage decreases in the following order: heparin > heparan sulfate > dermatan sulfate = chondroitin sulfates A and C > hyaluronic acid = K5 polysaccharide, indicating that both the degree of sulfation and the backbone structure of GAG modulate its interaction with bFGF. This is confirmed by the different capacity of various sulfated compounds (including dextran sulfates, suramin, trypan blue, and sulfate ion) to protect bFGF from tryptic digestion. Moreover, tryptic digestion studies performed with various heparin molecules and dextran sulfates of different size, ranging from 2.0 kDa to 500 kDa, indicate that the number of bFGF molecules which interact with a single molecule of polysaccharide is related to the molecular mass of the GAG and that six hexose residues are sufficient to protect 1-2 molecules bFGF. In conclusion, our findings indicate that the capacity of GAGs to protect bFGF from tryptic cleavage depends upon their size, sulfation, distribution of the anionic sites along the chain, and structural requirements of the bFGF molecule. These studies will help to design synthetic oligosaccharides endowed with different bFGF agonist and/or antagonist activities.

The NH2-terminal extension of high molecular weight forms of basic fibroblast growth factor (bFGF) is not essential for the binding of bFGF to nuclear chromatin in transfected NIH 3T3 cells.

Immunolocalization of basic fibroblast growth factor (bFGF) was investigated in NIH 3T3 cells transfected with a cDNA encoding for the 18 kD form of human bFGF (18 kD-bFGF) or with a bFGF cDNA encoding for both 18 kD-bFGF and NH2-terminal extended high molecular weight forms of bFGF (HMW-bFGFs). Nuclear and cytoplasmic bFGF-immunoreactivity was observed in both transfectants. Nuclear bFGF immunoreactivity was evenly distributed during interfase and associated with condensed chromosomes throughout the mitotic cycle. Cell fractionation, followed by Western blot analysis, confirmed the presence of 18 kD-bFGF and of HMW-bFGFs in the nucleus of transfected cells. Also, both 18-kD bFGF and HMW-bFGFs copurified with nuclear chromatin. After trypsin digestion, chromatin-bound bFGFs showed a rapid degradation of the nuclear-targeting NH2-terminal extension of HMW-bFGFs which were converted to the 18 kD form. On the contrary, 18 kD-bFGF appeared to be trypsin-resistant when bound to nuclear chromatin or to isolated eukaryote DNA. Thus, our data indicate that: i) both 18 kD-bFGF and HMW-bFGFs localize into the nucleus of transfected NIH 3T3 cells and bind to nuclear chromatin; ii) the interaction of all bFGF isoforms with nuclear chromatin is mediated by one or more sequences present within the 18 kD form; iii) the chromatin-binding domain of HMW-bFGFs is distinct from their nuclear-targeting domain.

Subcellular localization and biological activity of M(r) 18,000 basic fibroblast growth factor: site-directed mutagenesis of a putative nuclear translocation sequence.

Residues 27-31 (Lys-Asp-Pro-Lys-Arg) of the 155-amino acid form of basic fibroblast growth factor (bFGF) are in good agreement with a consensus sequence for nuclear translocation. To evaluate the role of this sequence in mediating the intracellular localization and biological activity of bFGF, basic residues Lys-27, Lys-30, and Arg-31 were changed to neutral glutamine residues by site-directed mutagenesis of the human bFGF cDNA. The bFGF mutant (M1Q-bFGF) was expressed in eukaryotic cells and in prokaryotic cells, from which it was purified to homogeneity. Transient expression of bFGF cDNA and of M1Q-bFGF cDNA in simian COS-1 cells followed by immunolocalization and by subcellular fractionation indicated that both molecules localize in the nucleus, as well as in the cytoplasm of transfected cells, and interact with nuclear chromatin and with eukaryote DNA in a similar manner. Prokaryotic expression of M1Q-bFGF cDNA yields a polypeptide endowed with a receptor-binding capacity and mitogenic activity similar to that exerted by wild-type bFGF. However, recombinant M1Q-bFGF showed a drastically reduced capacity to induce the production of urokinase-type plasminogen activator (uPA) in endothelial cells. The uPA-inducing activity of M1Q-bFGF was fully restored by the presence of soluble heparin in the culture medium. In conclusion, the sequence bFGF(27-31) does not appear to represent a nuclear translocation and/or retention sequence for bFGF. However, neutralization of its basic residues seems to modify the tertiary structure of the growth factor, thus affecting some of its biological properties.

Characterization of a biologically active extracellular domain of fibroblast growth factor receptor 1 expressed in Escherichia coli.

The functional features of a recombinant fibroblast growth factor (FGF) receptor (FGF-R) were investigated by expressing at high level in Escherichia coli a soluble non-glycosylated form of FGF-R1. The extracellular domain of the mature protein (XC-FGF-R), comprising the first 356 amino acids, was purified from a large-scale fermentation. After cell lysis, the protein was quantitatively found in the pellet. XC-FGF-R was solubilized using guanidine/HCl and allowed to refold using two dialysis steps. The refolded protein was obtained in a homogeneous form after ammonium sulphate precipitation and gel-filtration chromatography. The soluble receptor had the ability to form a complex with recombinant human basic FGF (rhbFGF) in solution, as demonstrated by immunoprecipitation with anti-(FGF-R) serum. Formation of a rhbFGF/XC-FGF-R complex was visualized by cross-linking experiments. Quantitative binding experiments with the XC-FGF-R immobilized on Affi-Gel resin showed high binding affinity for 125I-bFGF (Kd = 5-10 nM). Purified XC-FGF-R inhibited binding of 125I-bFGF to its high-affinity receptors on baby hamster kidney cells. These data suggest that glycosylation of the FGF-R is not necessary for its ligand-binding activity. The use of an E. coli expression system resulted in the efficient production of a soluble receptor in a form suitable for ligand/receptor structural studies and screening of new potential agonists and antagonists of angiogenesis. These results indicate that E. coli can be used for the production of complex molecules such as Ig-like receptors.

Basic fibroblast growth factor in Dupuytren's contracture.

Lesions excised from nine patients undergoing surgery for Dupuytren's contracture (DC) and three normal fascia were examined for the presence of the angiogenic protein basic fibroblast growth factor (basic FGF). Endothelial cell proliferation assays established basic FGF-like activity in extracts of DC. Western blotting confirmed the presence of an 18,000-dalton protein which was localized in the lesions by immunohistochemical staining. All of the cells implicated in the progression of the disease (endothelial cells, fibroblasts, and myofibroblasts) contain the growth factor. Endothelial cells within the narrowed or occluded vessels, as well as fibroblasts surrounding these vessels, stained intensely positive. In situ hybridization using an antisense probe for human basic FGF and its receptor's (FGFR-1) mRNA established the major difference between normal and DC tissues: their levels are significantly higher than in the normal tissues. Thus the cells in DC also express both basic FGF and FGFR-1, suggesting a potential autocrine/paracrine role for basic FGF in the pathogenesis of DC. This finding is thus the first description of a nontumoral proliferative disease that can be directly associated with increased basic FGF mRNA. The possibility that therapies can be developed on the basis that basic FGF and its receptor are expressed in DC is discussed.

Structure-function relationship of basic fibroblast growth factor: site-directed mutagenesis of a putative heparin-binding and receptor-binding region.

Basic residues Arg-118, Lys-119, Lys-128, and Arg-129 within a putative heparin-binding and receptor-binding region of the 155-amino acid form of basic fibroblast growth factor (bFGF) have been changed to neutral glutamine residues by site-directed mutagenesis of the human bFGF cDNA. The bFGF mutant (M6B-bFGF) was expressed in E. coli and purified to homogeneity. When compared to wild type bFGF, M6B-bFGF showed in cultured endothelial cells a similar receptor-binding capacity and mitogenic activity, but a reduced affinity for heparin-like low affinity binding sites, a reduced chemotactic activity, and a reduced capacity to induce the production of urokinase-type plasminogen activator. In vivo, M6B-bFGF lacked a significant angiogenic activity. Modifications of both the primary and the tertiary structure of bFGF appear to be responsible for the modified biological properties of M6B-bFGF, thus confirming the possibility to dissociate at the structural level some of the biological activities exerted by bFGF on endothelial cells.